LangChain+llama.cpp+llama-cpp-python工程学习笔记（一）

LangChain 是一个开发由语言模型驱动的应用程序的框架，有以下特性。

pip install langchain

报错ERROR: Could not find a version that satisfies the requirement langchain (from versions: none)ERROR: No matching distribution found for langchain

解决：pip install langchain -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com

langchain集成llama:

pip install llama-cpp-python -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com

报错:ERROR: Cannot unpack file C:\Users\96584\AppData\Local\Temp\pip-unpack-izfgtfwa\simple (downloaded from C:\Users\96584\AppData\Local\Temp\pip-req-build-1raavtqr, content-type: text/html; charset=utf-8); cannot detect archive format
ERROR: Cannot determine archive format of C:\Users\96584\AppData\Local\Temp\pip-req-build-1raavtqr

解决：换源

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn  llama-cpp-python

报错：ERROR: Could not build wheels for llama-cpp-python, which is required to install pyproject.toml-based projects   不识别nmake

解决：在powershell中更改编译器 在cmd中输入powershell

$env:CMAKE_GENERATOR = "MinGW Makefiles"
$env:CMAKE_ARGS = "-DLLAMA_OPENBLAS=on

-DCMAKE_C_COMPILER=C:/w64devkit/bin/gcc.exe

-DCMAKE_CXX_COMPILER=C:/w64devkit/bin/g++.exe"

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn  llama-cpp-python

报错：

:thread.c:(.text+0x103f): multiple definition of `pthread_self' ../../libllama.dll.a(d000850.o):(.text+0x0): first defined here collect2.exe: error: ld returned 1 exit status

解决：

在github上issue里有：添加：-DLLAVA_BUILD=OFF  取消LLAVA的支持

这是llama-cpp-python的bug

简单运行：

import llama_cpp

model = llama_cpp.Llama(
    model_path="D:/researchPJ/llamacpp/llama.cpp/models/llama-2-7b/ggml-model-q4_0.gguf",
)
print(model("The quick brown fox jumps ", stop=["."])["choices"][0]["text"])

报错：

RuntimeError: Failed to load shared library 'D:\code\langchain-llama\.venv\lib\site-packages\llama_cpp\libllama.dll': [WinError 193] %1 不是有效的 Win32 应用程序。

解决：

libllama.dll是32位的，需要重新编译为64位以适配windows

在cmake环境处添加-DCMAKE_GENERATOR_PLATFORM=x64以生成64位dll

报错：

MinGW Makefiles does not support platform specification, but platform x64 was specified.

解决：

创建文件如"mingw64.cmake"

其中内容：

set(CMAKE_GENERATOR_PLATFORM x64)

# Specify compilers

set(CMAKE_C_COMPILER C:/w64devkit/bin/gcc.exe)

set(CMAKE_CXX_COMPILER C:/w64devkit/bin/g++.exe)

更改编译环境：

 添加-DCMAKE_TOOLCHAIN_FILE=path/to/mingw64.cmake

重新编译：

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn llama-cpp-python --upgrade  --force --reinstall  --no-cache-dir
之后运行成功

create_chat_completion（）   return response['choices'][0]['message']['content']

langchain学习

from langchain_community.llms import LlamaCpp
llm = LlamaCpp(
    model_path="D:/researchPJ/llamacpp/llama.cpp/models/llama-2-7b/ggml-model-q4_0.gguf",
    temperature=0.75,
    n_gpu_layers=20,  # gpu accelerate
    n_threads=6,

)
prompt = """
what is langsmith?
"""
print(llm.invoke(prompt))

用模板来输入：

template = """Question: {question}

Answer: Let's work this out in a step by step way to be sure we have the right answer."""

prompt = PromptTemplate.from_template(template)

llm_chain = LLMChain(prompt=prompt, llm=llm)

question = "What NFL team won the Super Bowl in the year Justin Bieber was born?"
print(llm_chain.run(question))

提示+模型+输出解析：

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are world class technical documentation writer."),
    ("user", "{input}")
])

output_parser = StrOutputParser()

llm_chain = LLMChain(prompt=prompt, llm=llm, output_parser=output_parser)

print(llm_chain.invoke({"input": "how can langsmith help with testing?"}))

嵌入模型+向量存储+提示+模型+输出解析：

pip install beautifulsoup4

pip install faiss-cpu

from langchain.callbacks.manager import CallbackManager
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
from langchain.chains import LLMChain
from langchain.prompts import PromptTemplate
from langchain_community.llms import LlamaCpp
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_community.document_loaders import WebBaseLoader
from langchain.embeddings import LlamaCppEmbeddings
from langchain_community.vectorstores import FAISS
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.chains.combine_documents import create_stuff_documents_chain
from langchain.chains import create_retrieval_chain



model_path = "D:/researchPJ/llamacpp/llama.cpp/models/llama-2-7b/ggml-model-q4_0.gguf"

embeddings = LlamaCppEmbeddings(model_path=model_path)

llm = LlamaCpp(
    model_path=model_path,
    temperature=0.75,
    n_gpu_layers=20,  # gpu accelerate
    n_threads=6,
    n_ctx=2048,

)

loader = WebBaseLoader("https://docs.smith.langchain.com")

docs = loader.load()

text_splitter = RecursiveCharacterTextSplitter()
documents = text_splitter.split_documents(docs)
vector = FAISS.from_documents(documents, embeddings)

prompt = ChatPromptTemplate.from_template("""Answer the following question based only on the provided context:

<context>
{context}
</context>

Question: {input}""")

document_chain = create_stuff_documents_chain(llm, prompt)

retriever = vector.as_retriever()
retrieval_chain = create_retrieval_chain(retriever, document_chain)

response = retrieval_chain.invoke({"input": "how can langsmith help with testing?"})
print(response["answer"])

# LangSmith offers several features that can help with testing:...

加载要索引的数据。使用 WebBaseLoader

使用嵌入模型将文档摄取到向量存储

将创建一个检索链。 该链将接受一个传入的问题，查找相关文档，然后将这些文档与原始问题一起传递到 LLM 中，并要求它回答原始问题。